A homomorphic system to reduce speckle in videos

In this paper, we propose an unbiased homomorphic system to reduce speckle in videos. The speckle is modeled as a multiplicative noise having a lognormal distribution. In the proposed homomorphic system, first, the natural logarithm function is applied on the corrupted signal to transform the multiplicative nature of the noise into an additive one. Then, a spatial filter is applied to the signal obtained after the transformation. This is followed by the application of a temporal filter, which uses a change detection technique to measure the interframe motion, to the output of the spatial filter. The spatial estimate obtained from the spatial filter and the spatiotemporal estimate at the output of the temporal filter are then combined in a suitable way to get an intermediate signal. Exponentiation is then applied to this intermediate signal to get an estimate of the original signal. Then a technique to compensate for the biasing that occurs within the homomorphic system is applied to the output of the system in order to get an unbiased estimate of the original signal. The quantitative and the qualitative performance of the proposed system are studied and compared with that of the two existing schemes. It is found that the proposed filters have much less processing time and also perform better noise reduction compared to the other two.